Direct movement has long relocation time and overuses the redundant sensor Motivation Coverage under random deployment Coverage under clustering All the sensors are mobile Coverage hole exists? Move Solution: utilizing Mobile Sensor! Coverage Hole Detection Coverage hole detection: Only check local Voronoi cell Utilizing Voronoi diagram: Composed by the bisectors of neighbor sensors Each point in a Voronoi cell is closer to the sensor in this cell than any other sensors If the sensor can not detect the events happened in its Voronoi cell, no other sensors can Guiling Wang, Dr. Guohong Cao, and Dr. Tom La Porta (guiwang, gcao, gcao, direct the movement of sensors to increase coverage General idea: detecting coverage hole  move to heal the hole Calculate the target location (by VEC, VOR or Minimax) sensing range Coverage!! Sensor failure Deploying more static sensors cannot solve the problem due to wind or obstacles New event or new interest Redundant Sensor Discovery Organize sensors in one grid into a cluster Cluster head detects redundant sensors and maintains the information Movement Arrangement Redundant sensor Tradeoff between Sensor Cost and Coverage A mix of mobile and static sensors Mobile sensor: hole-healing server –Base price:the coverage hole generated by their leaving Static sensor: bidder –Bids: estimated size of the detected coverage hole Initialization Static sensor: –broadcast location Mobile sensor –base price = 0 Service advertisement Mobile sensor: Broadcast Bidding Static sensor: Send to the closest mobile sensor Bid > base price Serving Mobile sensor: Choose the highest bid Move Base price= accepted bid Base price increases monotonically and protocol stops naturally when no bidding message is flying Bidding Protocol Mobile percentage Algorithm tested All mobileVEC 10%~50%Bidding All staticRandom Money needed for certain Coverage (Static sensor: $1; Mobile sensor: $n) $ for 90% coverage $ for 98% coverage Sensor needed for certain Coverage Industry Day NET-CENTER.PSU Redundant sensor Sensor RelocationSensor Deployment Objectives: Least affect to applications (topology) Short relocation time Energy efficient and load-balanced Solution: Discover redundant sensors  only move the redundant sensors and cascading (relay) sensors if needed Supply quorum Demand quorum demand quorum: grids in a column supply quorum: grids in a row need sensor: search demand quorum have redundant sensor: notify supply quorum Why cascading movement? Which cascading schedule? Much more energy consumption Good! Strategy: make a cascading schedule  move following the schedule (movement is more costly than communication!) VEC Motivated by the attributes of electrical particles Virtual force pushes sensors away from dense area B C A VOR Move towards the farthest Voronoi vertex Avoid moving oscillation: stop for one round if move backwards B M B C A B M Minimax Move to where the distance to the farthest voronoi vertex is minimized B M N B M N Why mixed: tradeoff between coverage and sensor cost Goals: satisfy the requirement of relocation delay minimize the difference between total energy consumption and minimum remaining energy Approach: distributed implementation of dynamic programming Become strategic area GridQuorum system Guiling Wang, Guohong Cao, and Tom La Porta, “Movement- Assisted Sensor Deployment”, IEEE INFOCOM’04, March Guiling Wang, Guohong Cao, and Tom La Porta, “A Bidding Protocol for Sensor Deployment”, IEEE ICNP’03, Nov., Guiling Wang, Guohong Cao, and Tom La Porta, “Sensor Relocation in Mobile Sensor Networks”, in submission. Movement-Assisted Sensor Deployment and Sensor Relocation